Sulfate-based coagulants can suppress methanogenesis in treated oil sands fine tailings

Bitumen extraction from mined oil sands ore generates a large volume of fluid fines tailings (FFT) that must be incorporated into either aquatic or terrestrial reclamation landforms. Mine operators are developing various tailings technologies to accelerate FFT dewatering, including the addition of chemical coagulants and flocculants. However, the impacts of these coagulants and flocculants on biogeochemical processes in treated FFT are not fully understood. We conducted anaerobic batch experiments to examine the influence of different doses (i.e., 0, 500, 1000, and 1500 ppm) of sulfate-based coagulants, including aluminum sulfate (alum) [Al₂(SO₄)₃∙nH₂O], ferric sulfate (ferric) [Fe₂(SO₄)₃∙nH₂O], and calcium sulfate (gypsum) [CaSO₄∙2H₂O], on biogenic gas production and microbial communities in treated FFT. Our results show that sulfate addition stimulated microbial sulfate reduction, which inhibited methanogenesis in coagulated FFT relative to experimental controls. Sulfate depletion preceded increased methane production in the 500 ppm gypsum experiment, while larger ferric and alum doses produced higher sulfate concentrations and larger pH decreases. 16 S rRNA sequencing revealed that Comamonadaceae, Anaerolineaceae, and Desulfocapsaceae were the major bacterial families, while Methanoregulaceae and Methanosaetaceae dominated the archaeal families in all treatments. Precipitation of iron(II) sulfides limited dissolved hydrogen sulfide concentrations in experiments where Fe availability was not limited. Our results indicate that addition of sulfate-based coagulants can stimulate microbial sulfate reduction and suppress methanogenesis. However, resumption of methane production following sulfate depletion reveals complex interactions among biogeochemical reaction pathways. Overall, this study demonstrates that biogeochemical cycling of carbon, sulfur, and iron are important considerations for the development and implementation of tailings treatment technologies.

Fluid fine tailings (FFT) management and reclamation are considerable challenges for the oil sands industry in northern Alberta, Canada. The combined physical and chemical characteristics of FFT solids and associated water inhibits aggregation, thereby slowing settlement and dewatering in tailings ponds [24]. Additionally, residual hydrocarbons within FFT deposits support diverse anaerobic microbial metabolisms including methanogenesis, sulfate reduction, and iron(III) reduction [75, 79, 80]. The slow settlement and dewatering behavior drive growth of FFT inventories in tailings ponds, and regulations have been implemented to mitigate tailings growth and progressively reclaim mined-out areas (ERCB 25; Government of Alberta 33; AER 1. To curb the ongoing inventory growth, oil sands operators have been developing new tailings treatment technologies. Some of these technologies use chemical coagulants, including aluminum sulfate [Al₂(SO₄)₃∙nH₂O], ferric sulfate [Fe₂(SO₄)₃∙nH₂O], and calcium sulfate [CaSO₄∙2H₂O], to accelerate FTT settlement and dewatering.

Sulfate-based coagulants contribute sulfate (SO₄²⁻) and associated cations (i.e., Al³⁺, Fe³⁺, Ca²⁺) to treated FFT. This sulfate and iron(III) may serve as electron acceptor for microbial sulfate reduction and iron reduction, respectively, while residual hydrocarbons also support methanogenesis [28, 72, 73, 80]. Complex organics are biodegraded by syntrophs to produce substrates such as acetate and H₂, which support other microbial metabolisms [28]. [36] reported that residual bitumen is a poor substrate for microbial growth owing to a high proportion of recalcitrant hydrocarbons. However, unrecovered naphtha diluent added during bitumen froth treatment can support methanogenesis, sulfate reduction, and other anaerobic microbial metabolisms [28]. Iron-reducing bacteria (FeRB) and sulfate-reducing bacteria (SRB) produce Fe²⁺ and H₂S, respectively, while methanogens can generate CH₄ via acetoclastic or hydrogenotrophic pathways [56].

Interactions among microbial metabolisms play a crucial role in the complex relationships governing iron reduction, sulfate reduction, and methanogenesis [13, 63, 78,79,80]. These biogeochemical redox processes can occur simultaneously under circumneutral pH conditions where substrates are readily available [13, 78, 80]. SRB can access more usable energy in the near-neutral pH range where sulfate and organic carbon are abundant [13]. FeRB can enhance sulfate reduction by producing Fe²⁺ that reacts with H₂S to form Fe(II) sulfide minerals [63]. This reaction scavenges dissolved H₂S, which is toxic to SRB at high concentrations, thereby enhancing sulfate reduction [75]. Furthermore, sulfate reduction may stimulate methanogenesis in organic carbon–limited environments by oxidizing acetate to CO₂, which can then be used by hydrogenotrophic methanogens [56, 79]. Furthermore, CH₄ can serve as both growth substrate and electron donor for iron(III) reduction mediated by FeRB [6, 45]. These complex microbial interactions strongly influence biogeochemical cycling of carbon, sulfur, and iron in FFT deposits.

Here, we examine biogeochemical responses to the addition of sulfate-based coagulants to oil sands FFT. Our research objectives were to assess changes in biogenic gas production and microbial community compositions in FFT treated with varied doses of aluminum sulfate (alum), ferric sulfate (ferric), and calcium sulfate (gypsum). We performed anoxic laboratory batch experiments and monitored pore-water chemistry, gas production, and microbial communities over time. Our findings offer insight into the complex dynamics among biogeochemical redox processes within treated tailings and can support ongoing development of treatment technologies for oil sands FFT.

Fluid fine tailings

Laboratory batch experiments used FFT collected from MLSB, which is located at the Syncrude Mildred Lake operation in northern Alberta, Canada. Tailings stored in this ~ 10 km² facility contain unrecovered naphtha diluent used during bitumen froth treatment [28, 56, 68, 69, 75]. Two 20 L samples were collected from approximately 2 m below the tailings-water interface following the fluid sampler method described by [24]. Pore-water exhibited circumneutral pH and elevated concentrations of Na, Cl, Ca, Mg, HCO₃ and NH_3, characteristic of FFT deposits [24, 36]. Gravimetric moisture contents were determined by oven drying at 105 °C for 72 h, and solid contents were 65% (± 4) and 35% (± 4), respectively.

Coagulants

Aluminum octadecahydrate [Al₂(SO₄)₃∙18H₂O], calcium sulfate dihydrate [CaSO₄^.2H₂O], and ferric sulfate pentahydrate [Fe₂(SO₄)₃^.5H₂O] (ACS Reagent Grade, Sigma Millipore, Canada) were used for coagulation. Alum and ferric were prepared as acidic stock solutions with pH adjusted to 2.1 and 1.0, respectively, to prevent Al(III) and Fe(III) hydrolysis and subsequent (oxy)hydroxide precipitation. The initial alum stock contained 20,000 ppm Al³⁺ and 106,800 ppm SO₄²⁻, while the concentrations of the ferric sulfate stock solution were 20,000 ppm Fe³⁺ and 51,600 ppm SO₄²⁻. These concentrations were chosen to minimize the amount of coagulant solution added. A range of initial Al, Fe, and Ca concentrations, which include 0 (control), 500, 1000, and 1500 ppm (Table 1), were chosen based on concentrations that have been reported to be used in coagulation processes [50, 58, 81, 87, 88]. The doses and tailings volumes were calculated and added based on the initial FFT pore-water content (see supplementary content: A1).

Batch experiments

Batch experiments (Table 1) were established in an anaerobic chamber (≤ 2 vol% H_2(g), balance N_2(g); Coy Laboratory Products, USA). All glassware, serum bottles, and centrifuge tubes used for these experiments were soaked in a 5% nitric acid bath for 24 h and rinsed three times with ultrapure water prior to use. The FFT was thoroughly mixed with a paint mixer and 100 g aliquots of the bulk mass were added to 120 mL amber glass serum bottles for all the controls (0 ppm) and 95 ± 2.5 g for the remaining doses (500, 1000, and 1500 ppm). Samples were prepared in triplicate under ambient laboratory conditions for each coagulant dose, resulting in 324 individual bottles. The bottles were immediately transferred into an anaerobic chamber, where the coagulants were thoroughly mixed into the FFT. The bottles were loosely covered with parafilm to limit evaporation and allowed to equilibrate with the anaerobic atmosphere. After 24 h, the bottles were capped with blue bromobutyl rubber stoppers and crimp sealed. The bottles stirred twice weekly using a vortex mixer, including once 24 h before sampling. Headspace gases were vented regularly to maintain atmospheric pressure.

Gas sampling and analysis

Headspace gas samples were collected and analyzed over time (Table 1) using 5 cc gas-tight syringes (Series A-2, VICI Precision Sampling, USA). The headspace gas samples were analyzed using a Trace 1310 gas chromatography (GC) system (Thermo Scientific, USA) equipped with a thermal conductivity detector (TCD) and a flame ionization detector (FID). Analyses were performed using ultrahigh-purity He_(g) (99.999%, Praxair, Canada), and calibration curves were generated using certified reference gas mixtures (CalgasWarehouse, GASCO, USA). Dissolved gas concentrations were calculated from headspace concentrations using temperatures, pressures, volumes, and established constants [5, 38]. Additional details on gas sampling and analysis are provided in the Supplementary Material.

Pore-water sampling and analysis

Destructive sampling was performed at pre-determined intervals (0, 1, 2, 4, 8, 16, 32, and 49 weeks) for water and solids. During sampling, each bottle was opened and thoroughly stirred on a vortex mixer for 10 s before the FFT was transferred into two 50 mL polypropylene conical centrifuge tubes; these were capped and sealed in a centrifuge rotor within the anaerobic chamber. The rotor was then removed from the chamber and the tubes were centrifuged (Sorvall X Pro Series, Thermo Scientific, USA) at 13,800g for 30 min. The rotor was returned to the glovebox, opened, and supernatant was collected for geochemical analyses.

Reduction-oxidation potential (Eh), pH, and EC were measured using a redox electrode, pH electrode, and EC cell, respectively. Samples were filtered through polyethersulfone (PES) syringe filter membranes (Minisart, Sartorius AG, Germany) and 30 mL syringes (Henke-Sass Wolf GmbH, Germany). The pH electrode (Orion 8156NUWP ROSS Ultra, Thermo Scientific, USA) was calibrated and recalibrated at intervals before and during measurements using NIST-traceable pH 4, 7, and 10 buffer solutions. The Eh electrode (Orion 9678BNWP, Thermo Scientific, USA Scientific) was calibrated before measurements using ZoBell’s [55, 95] and Light’s [46] solutions, and measured Eh values were corrected to standard hydrogen electrodes. The EC electrode was calibrated before measurements using a standard 1413 µS cm⁻¹ NaCl solution (Thermo Scientific, USA). Alkalinity was measured by titrating 5 mL of filtered samples to the bromocresol green methyl red (Ricca Chemical Company, USA) endpoint using either 1.6 N–0.16 N H₂SO₄ based on Hatch Method 8203. Dissolved ƩH₂S (ƩH₂S = H₂S + HS^-) and ammonia (ƩNH₃-N = NH₃ + NH₄⁺) were measured by spectrophotometry (Model DR2800, Hach Chemical Company, USA) using the methylene blue method [47] and the salicylate method [39], respectively. Theoretical Eh values were calculated using the Nernst Equation based on HS⁻/SO₄²⁻ and CH₄/CO₂ redox couples present in the samples.

Trace elements in solution were measured using inductively coupled plasma–mass spectrometry (ICP-MS; Thermo Fisher iCAP TQE; EPA Method 200.8, USA) and cations were measured using inductively coupled plasma–optical emission spectroscopy (ICP-OES; SPECTROBLUE, Spectro analytical instruments, Germany; EPA Method 200.7). Inorganic anions were measured using ion chromatography (IC; ICS2100; Dionex Corporation Thermo Fisher Scientific, USA). Samples were passed through 0.1 μm polyethersulfone (PES) syringe filter membranes, and ICP-MS and ICP-OES samples were acidified to pH < 2 using trace metal–grade nitric acid (HNO₃; OmniTrace, EMD Millipore, USA). All water samples were stored in high-density polyethylene (HDPE) bottles at 4 °C pending analyses.

DNA extraction and quantification

Deoxyribonucleic acid (DNA) extractions were performed in triplicate on FFT samples using a commercially available FastDNA SPIN extraction kit for soil (MP Biomedicals, CA, USA). Frozen samples stored in sterilized 15 mL conical polyethylene centrifuge tubes were thawed in an anaerobic chamber, and DNA extraction was done following the protocol suggested by the manufacturer. The extracted DNA was further purified using DNeasy^® PowerClean^® Pro Cleanup Kit (QIAGEN Group, Germany) to remove any lingering impurities and to improve the DNA yields. The purified DNA was quantified using the Qubit High Sensitivity dsDNA Assay Kit (Life Technologies, CA, USA) and readings were taken on a Qubit 2.0 Fluorometer (Invitrogen, Life Technologies, CA, USA) at the geomicrobiology laboratory of Dr Joyce McBeth at the University of Saskatchewan, in accordance with the protocols provided by the manufacturer. The extracted DNA samples were subsequently stored at − 20 °C prior to sequencing. Additional details on DNA extraction and purification are provided in the Supplementary Material.

DNA sequencing

Microbial communities in both treated and untreated FFT were assessed by 16 S ribosomal ribonucleic acid (rRNA) gene amplicon sequencing. Triplicate DNA samples were amplified by PCR for sequencing in a one-step process with Illumina adapters attached to primers. Universal primers 15FB/806RB [86] were employed to target the V4 region of the 16 S rRNA gene in both Bacteria and Archaea. An average of 15,000 paired reads were obtained per sample (15,000 forward and corresponding 15,000 reverse reads). Additional details on DNA sequencing are provided in the Supplementary Material.

Bioinformatics and statistical analyses

The raw MiSeq reads were demultiplexed, pre-processed, and analyzed using Divisive Amplicon Denoising Algorithm 2 (DADA2) v1.8 software package [15] integrated within QIIME 2 version 2021 (https://qiime2.org/, [16]. The demultiplexed data were processed to remove primers and truncate forward and reverse reads at lower quality thresholds (approximately 225 forward and 175 reverse). Illumina sequencing error modeling and correction were done, followed by the assembly of paired reads into identical sequence units. More abundant parent amplified sequence variants (ASVs) were generated to eliminate chimeric ASVs, and an ASV table was constructed for downstream analyses (BioStudies, Accession: S‑BSST1350, DOI: https://doi.org/10.6019/S-BSST1350, Link: https://www.ebi.ac.uk/biostudies/studies/S-BSST1350?key=f98201dc-b766-479a-b654-ac59f9d780df). Sequences were clustered into operational taxonomic units (OTUs), which were assigned to representative sequences using the naïve Bayesian classifier implemented in QIIME 2 with scikit-learn (v.0.21.3) trained against SILVA release 138 clustered at 99% similarity for 16 S rRNA genes. Assignments were accepted with a confidence threshold of 0.7 or higher.

Statistical comparisons between the abundance values were performed using one-way ANOVA (Tukey’s test) and visualizations were created using the R package ggplot [90] code is available at https://git.io/fptlp). Samples with less than 5000 reads were excluded prior to the statistical analyses to minimize errors associated with low sequencing depth and to enhance the accuracy of the diversity estimates [76]. Alpha (α) diversity (variation in microbial taxa within sample) was generated based on the rarefied OTU table by using the following alpha metrics — Observed Species, Chao1, Shannon, and Simpson calculated via QIIME [16]. Observed species and Chao1 metrics estimate species richness, while Shannon and Simpson indices measure diversity within each sample. All analyses were performed using the MicrobiomeAnalyst cloud platform with the MicrobiomeAnalystR packages [23]. Shifts in microbial communities, in response to different coagulants addition and doses over time, were visualized using bar charts and heatmaps.

Effects of chemical treatments on pore-water chemistry

pH, eh, alkalinity, and electrical conductivity

The pore-water pH in controls ranged between 8.00 ± 0.01 and 8.51 ± 0.02 from week 0 to week 49, which falls within the pH values previously reported for FFT [4, 18, 24, 58]. The pH values slightly increased from week 0 to week 2 in all coagulant treatments and generally plateaued with slight variability after week 4 (Fig. 1). Higher alum and ferric sulfate doses resulted in lower pH. The pH values for the 500 ppm alum treatment ranged from 6.42 ± 0.03 to 7.10 ± 0.01 and were consistently < 6.0 at higher doses (1000 and 1500 ppm). Ferric sulfate treatments had less impact on pH than alum, and varied between 6.28 ± 0.06 and 7.68 ± 0.06 in all ferric sulfate treatments. Gypsum amendments slightly decreased the solution pH, with values ranging between 7.41 ± 0.02 and 8.03 ± 0.01 in all gypsum treatments.

Measured redox potential (Eh) values, which ranged from 830 ± 3.75 to 595 ± 13.3 mV in the controls and varied between 750 ± 7.91 and 413 ± 11.2 mV in all treated batches, were inconsistent with observed chemistry (Fig. 1). The calculated theoretical Eh values for HS⁻/SO₄²⁻ ranged from − 33 to − 306 mV while CH₄/CO₂ values ranged between − 55 and − 310 mV in all batches.

Alkalinity (reported as mg L⁻¹ of CaCO₃) in the controls ranged between 740 and 1140 mg L⁻¹ throughout the experiment (Fig. 1), generally consistent with the alkalinity values reported for FFT pore water [4, 36, 44, 72]. Alkalinity concentrations generally decreased with increasing coagulant doses. The observed trends in all doses indicate that alkalinity concentrations increased from week 0 to week 8 and were slightly variable from week 16 to week 49, especially in the alum and ferric sulfate treatments. The trends observed in gypsum treatments were less consistent from week 16 to week 49. Alkalinity concentrations were similar in all 500 ppm coagulant doses with values ranging between 320 and 920 mg L⁻¹, while alkalinity in the 1000 and 1500 ppm treatments was lower in the alum treatments (40–220 mg L⁻¹) than in treatments with similar doses of ferric sulfate (280–700 mg L⁻¹) and gypsum (400–800 mg L⁻¹).

Pore-water pH, alkalinity, and electrical conductivity for different doses (0, 500, 1000, and 1500 ppm) of coagulants (alum, ferric sulfate, and gypsum) from week 0 to week 49. Where visible, the error bars represent 1σ

Full size image

The EC ranged from 1.94 to 7.26 mS cm⁻¹ in the controls, which is higher than the average values reported for FFT [4, 18, 24, 63]. The observed trends were highly variable, especially in gypsum treatments, and EC increased with increasing coagulant doses throughout the experiments (Fig. 1). The measured EC concentrations ranged from 2.31 ± 0.00 to 10.7 ± 0.03 mS cm⁻¹ in all batches.

Dissolved iron, sulfur, and nitrogen

The dissolved total Fe concentration in the pore water of the control batches was < 1.0 mg L⁻¹, which is consistent with low concentrations previously reported in FFT [24, 56, 87]. The dissolved Fe concentrations increased with increasing amendment rates (Fig. 2). Total Fe concentrations in the 500 ppm alum batch consistently decreased from week 0 (187 ± 7.40 mg L⁻¹) to week 49 (< 6.00 mg L⁻¹), while the concentrations for 1000 and 1500 ppm alum were higher and did not decrease over time (Fig. 2). Iron concentrations in ferric sulfate treatments consistently decreased from week 0 to week 49 and were mostly below detection for gypsum treatments.

Sulfate concentrations varied widely in controls, ranging between 143 ± 6.96 and 3.26 ± 0.19 mg L⁻¹ over the course of the experiment, which is within the wide range previously reported for FFT [4, 18, 36, 56, 78]. All treatments exhibited positive trends between SO₄²⁻ concentrations and amendment rates (Fig. 2). The sulfate concentration increased in the 500 ppm alum treatments from week 0 to week 4 and decreased from 2810 ± 107 (week 4) to 1420 ± 157 mg L⁻¹ (week 49); sulfate was consistent from week 1 to week 49 in the 1000 ppm alum (4750 ± 181 mg L⁻¹) and 1500 ppm alum (6840 ± 40.6 mg L⁻¹) batches. The SO₄²⁻ concentrations in the 500 ppm ferric sulfate treatments increased from week 0 (1370 ± 22.2) to week 1 (1610 ± 137 mg L⁻¹) and decreased sharply toward week 49 with concentrations ranging between 5.40 ± 0.34 (week 16) and 25.86 ± 0.46 mg L⁻¹ (week 49). The dissolved SO₄²⁻ concentrations in the 1000 and 1500 ppm ferric sulfate treatments ranged between 3890 ± 89.8 and 1850 ± 34.7 mg L⁻¹ throughout the experiment. The 500 ppm gypsum treatments exhibited a decrease in SO₄²⁻ concentrations from week 0 (1530 ± 25.7) to week 49 (230 ± 9.06 mg L⁻¹), while the 1000 and 1500 ppm gypsum treatments ranged between 1580 ± 78.3 and 3530 ± 15.0 mg L⁻¹ over the course of the experiment.

Dissolved concentrations of total Fe, SO₄²⁻, ΣH₂S, and ƩNH₃-N in FFT pore water for 0, 500, 1000, and 1500 ppm of alum (Al³⁺), ferric sulfate (Fe³⁺), and gypsum (Ca²⁺) from week 0 to week 49. Where visible, the error bars represent 1 σ

Full size image

Total H₂S concentrations in the controls varied between 21.3 ± 0.58 and 74.0 ± 13.1 µg L⁻¹. Generally, there were inverse trends between ΣH₂S concentrations and increasing alum and ferric sulfate treatments, while gypsum doses displayed positive trends (Fig. 2). The dissolved total H₂S concentrations in the 500 ppm alum treatment varied between < 6.00 and 52.0 ± 7.00 µg L⁻¹, while concentrations in the 1000 and 1500 ppm alum batches ranged from < 1.00 to 33.5 µg L⁻¹ over time. The ΣH₂S concentrations in the 500 ppm ferric sulfate batch increased from week 0 (13.0 ± 3.18) to week 4 (35.7 ± 4.50 µg L⁻¹) and decreased toward week 49 with concentrations ranging between 1.33 ± 0.58 µg L⁻¹ (week 32) and 5.67 ± 2.08 µg L⁻¹ (week 49). Total H₂S concentrations in the 1000 and 1500 ppm ferric sulfate treatments were similar and ranged between 2.67 ± 00 and 43.7 ± 1.53 µg L⁻¹. The ΣH₂S concentrations in the gypsum treatments did not follow a consistent trend before week 16, but ΣH₂S concentrations increased from 62.7 ± 2.31 (week 16) to 232 ± 4.51 µg L⁻¹ (week 49) in 1500 ppm treatment (Fig. 2).

Dissolved ƩNH₃-N concentrations ranged between 1.19 ± 0.03 and 9.44 ± 0.07 mg L⁻¹ in controls, which falls within the range previously reported for FFT [24, 36, 78]. There were no clearly defined trends between NH₃-N and coagulant amendment rates. The dissolved NH₃-N concentrations in the 500 and 1000 ppm alum batches varied between 13.5 ± 0.83 and 20.2 ± 0.38 mg L⁻¹, while the NH₃-N concentrations in the 1500 ppm alum batch decreased from 21.2 ± 0.26 (week 0) to 1.31 ± 0.16 mg L⁻¹ (week 4) and then stabilized toward the end of the experiment (Fig. 2). The dissolved NH₃-N concentrations in the ferric sulfate treatments varied between 2.77 ± 0.09 and 17.0 ± 4.52 mg L⁻¹ over time. All gypsum-treated batches exhibited consistent NH₃-N concentrations ranging between 1.19 ± 0.03 and 9.44 ± 0.07 mg L⁻¹ throughout the experiment (Fig. 2). Nitrate (NO₃⁻) and nitrite (NO₂⁻) concentrations were consistently below method detection limits (MDLs) of approximately 0.23 and 0.015 mg L⁻¹, respectively, in all samples.

Dissolved carbon dioxide and methane

Dissolved CO₂ concentrations in all treatments increased with increasing coagulant dose. These concentrations greatly increased from week 0 to week 2 and remained consistent with slight variability from week 8 to week 49 (Fig. 3), especially in alum and ferric sulfate treatments. In the controls, pore-water concentrations of CO₂ ranged between 4.19 ± 0.80 and 49.0 ± 1.36 mg L⁻¹ over the course of the experiment. In the 500 ppm alum batch, CO₂ concentrations increased from 130 ± 40.1 mg L⁻¹ at week 0 to 621 ± 30.2 mg L⁻¹ at week 2 and became relatively stable from week 8 (801 ± 14.1 mg L⁻¹) to week 49 (908 ± 58.2 mg L⁻¹). The CO₂ concentrations in the 1000 and 1500 ppm alum batches followed similar patterns and varied between 276 ± 50.2 and 1750 ± 133 mg L⁻¹. The dissolved CO₂ concentrations for ferric sulfate treatments have similar trends with alum batches, but with relatively lower values (Fig. 3). For the 500 ppm ferric sulfate batch, CO₂ concentrations increased from week 0 (222 ± 121 mg L⁻¹) to week 2 (284 ± 50.2 mg L⁻¹) and were consistent from week 16 (272 ± 37.2 mg L⁻¹) to week 49 (266 ± 47.3 mg L⁻¹). Higher ferric sulfate doses (1000 and 1500 ppm) exhibited the same trend, with concentrations ranging between 268 ± 150 and 799 ± 48.5 mg L⁻¹ from week 0 to week 49. Dissolved CO₂ concentrations in the gypsum treatments followed the same general trends but were relatively low compared with the concentrations in the alum and ferric sulfate treatments (Fig. 3). For the 500 ppm gypsum treatment, CO₂ concentrations increased from week 0 (11.0 ± 3.33 mg L⁻¹) to week 16 (91.7 ± 1.21 mg L⁻¹) and stabilized from week 32 (82.2 ± 2.75 mg L⁻¹) to week 49 (85.3 ± 2.36 mg L⁻¹). The dissolved concentrations for the 1000 and 1500 ppm gypsum batches varied between 23.3 ± 7.75 and 167 ± 0.96 mg L⁻¹ over time.

Dissolved concentrations of CO₂ and CH₄ in FFT pore water for 0, 500, 1000, and 1500 ppm of alum (Al³⁺), ferric sulfate (Fe³⁺), and gypsum (Ca²⁺) from week 0 to week 49. Where visible, the error bars represent 1 σ

Full size image

Dissolved CH₄ concentrations for all the coagulant treatments, including the controls, were below detection at week 0 (Fig. 3). Within 4 weeks, CH₄ concentrations in the controls of alum batches increased from 0.21 ± 0.14 mg L⁻¹ (week 1) to 0.61 ± 0.01 mg L⁻¹ (week 4), then increased above 1 mg L⁻¹ at week 8 and reached 8.65 ± 0.15 mg L⁻¹ at week 49 (Fig. 3). Other alum-treated batches exhibited low dissolved CH₄ concentrations ranging between 0.14 and 0.45 mg L⁻¹ from week 1 to week 49. The CH₄ concentrations in the control of the ferric sulfate batches ranged between 0.23 ± 0.01 at week 1 and 0.5 ± 0.03 mg L⁻¹ at week 4, then increased to > 2.00 mg L⁻¹ at week 8 and further to 12.9 ± 0.33 mg L⁻¹ at week 49. For the 500 ppm ferric sulfate batch, dissolved CH₄ concentrations were low from week 1 to week 8 but increased from 0.51 ± 0.40 mg L⁻¹ at week 16 to 10.6 ± 0.67 mg L⁻¹ at week 49 (Fig. 3). Other ferric sulfate batches (1000 and 1500 ppm) had low CH₄ concentrations, which ranged between 0.11 and 0.35 mg L⁻¹ from week 1 to week 49. Dissolved CH₄ concentrations for the controls in gypsum batches ranged from 0.16 ± 0.05 mg L⁻¹ at week 1 to 0.76 ± 0.01 mg L⁻¹ at week 8 and increased from 2.21 ± 0.02 mg L⁻¹ (week 16) to 5.58 ± 0.02 mg L⁻¹ (week 49). The dissolved concentrations for gypsum treatments ranged between 0.01 mg L⁻¹ and 0.19 mg L⁻¹ from week 1 to week 49 (Fig. 3). The dissolved CH₄ concentrations in the controls were consistent with the ranges reported below the tailings-water interface (TWI) in the field [30, 56].

Microbial community

Diversity and richness

Microbial diversity within the samples exhibited variation based on coagulants used, though no consistent trends were apparent across all treatments (Fig. 4). Alpha-diversity values showed similarities between observed species and Chao1 index; however, the Shannon and Inverse Simpson indices remained relatively low across all treatments. In the alum treatments, the number of observed species (based on unique OTU counts) and the Chao1 index estimated that species richness fluctuated across batches; and the measure of species abundance, evenness, and dominance using Shannon and Inverse Simpson indices also varied in all batches. The highest observed species and diversity occurred in the 1500 ppm alum treatment (week 16), and the lowest values in the 500 ppm alum treatment (week 49). For ferric sulfate-treated samples, similar temporal variability was noted, with the highest species richness and diversity observed in 1500 ppm treatment (week 16), and the lowest in the 500 ppm treatment (week 49). Notably, gypsum-treated batches at 500 and 1000 ppm showed increased species richness and diversity, while these metrics declined at 1500 ppm. Shannon and Inverse Simpson indices remained relatively stable in gypsum treatments over time.

Stacked column plot of Observed Species, Chao1, Shannon, and Inverse Simpson indices values for different concentrations of alum, ferric sulfate, and gypsum from week 0 to week 49. The first numbers (0, 16 and 49) represent the number of weeks of the experiment, while the numbers after the hyphen (0.0, 0.5, 1.0, and 1.5) represent coagulant doses (0, 500, 1000, and 1500 ppm)

Full size image

Bacteria

Bacteria dominated the microbial community in both treated FFT and controls. The Proteobacteria phylum exhibited the highest relative abundance, comprising 23–58% for controls and 21–33% for treated FFT. The phylum Desulfobacterota accounted for 8–42% of all sequence reads, while other notable phyla included Chloroflexi (7–27%), Firmicutes (2–17%), and Latescibacteria (1–11%). Additionally, a substantial proportion of unclassified Bacteria were also present with relative abundances exhibiting considerable variability among treatments. Sequence reads for unclassified Bacteria varied from 20 to 54% for alum treatments, 9–25% for ferric treatments, and 21–24% for gypsum treatments (Fig. 5).

Sequence reads associated with the Proteobacteria phylum were dominated by taxa from the Gammaproteobacteria class, including Burkholderiales, Cellvibrionales, and Immundisolibacterales. The most abundant families classified to Burkholderiales were Comamonadaceae, Rhodocyclaceae, and Hydrogenophilaceae. Comamonadaceae dominated this order with initial relative abundance between 15% and 51% (controls) at week 0 and between 4% and 12% at week 49. Coagulant additions resulted in decreased abundance in all batches compared with the controls (Fig. 5). Rhodocyclaceae, which are involved in methanogenic hydrocarbon degradation and contain H₂-consuming microbes associated with Hydrogenophilaceae [7, 28] were present at 1–8% relative abundance in all treatments. Families classified to Cellvibrionales and Immundisolibacterales were Porticoccaceae (1–11% abundance) and Immundisolibacteraceae (0.5–4%), respectively.

The dominant classes within the Desulfobacterota included Desulfobulbia, Desulfuromonadia, Syntrophia, and Desulfobacteria. These classes are diverse and include sequences for putative sulfate reducers, iron reducers, and methanogenic hydrocarbon degraders, as reported in previous studies [7, 10, 28, 42, 71]. Families classified under the Desulfobulbia class, in order of abundance in the alum and ferric sulfate treatments, were Desulfocapsaceae, Desulfurivibrionaceae, and Desulfobulbabeae. Gypsum treatments did not display an abundance of any family (Fig. 5).

Bubble chart representing the percentage relative abundance of the 18 most abundant bacterial families. Each bubble represents % reads relative to all the major families in each sample. The first numbers (0, 16, and 49) represent the number of weeks, while the numbers after the hyphen (0.0, 0.5, 1.0, and 1.5) represent coagulant doses (0, 500, 1000, and 1500 ppm). R.A = Relative Abundance

Full size image

The major Desulforomonadia family was Geothermobacteraceae, which is associated with iron reduction, was present in low abundance (0–2%) in alum and ferric sulfate treatments but was not detected in gypsum treatments. Families in Syntrophia included Smithelaceae and uncultured organisms; these were identified at low abundance in alum and ferric sulfate treatments (0–4%) and were mostly not present in gypsum treatments. Within the Desulfobacteria class were the Desulfobacterales and Desulfatiglandales orders and the families Desulfosarcinaceae and Desulfatiglandaceae. In the alum treatments, the Desulfosarcinaceae family was present at the highest abundance (23%) in the 500 ppm alum batch at week 49, while the abundance in other alum batches varied between 0.5 and 2% over time. The abundance of this family in the ferric sulfate treatments was highest in the 1000 ppm batch at week 16 (24%) but decreased at week 49 (17%). The abundance of this family in gypsum-amended batches (8–12%) was higher than in the controls, similar to the trend observed with other coagulants. The Desulfatiglandaceae family was mostly not present in the alum treatments and ranged between 1 and 3% in the ferric sulfate and gypsum treatments.

Archaea

Archaeal taxa were identified in all samples, with most sequences associated with the phylum Halobacterota (Fig. 6). The relative abundance of phylum Halobacterota varied significantly, ranging from 1 to 50% in both untreated and treated samples. Other archaeal phyla, which included Euryarchaeota (1–18%) and Thermoplasmatota (1–11%), were also present (Fig. 6). The Halobacterota phylum, which dominated the Archaeal communities, included the classes, Methanomicrobia, Methanosarcinia, and various uncultured classes. The Methanomicrobia class consisted of the families Methanomicrobiaceae and Methanoregulaceae, which included genera Methanoregula, Methanolinea, and Methanoculleus. These families collectively accounted for 3–46% of the total archaeal sequences across all samples (Fig. 6). The Methanosaetaceae family exhibited high relative abundances, contributing 24–41% of the archaeal sequences. Similar taxa have been reported in MLSB [56] and in laboratory analyses of FFT samples from other active tailings ponds [78, 79]. The uncultured Halobacterota taxa identified in the samples appeared at a low abundance (0–9%) across all treatments. Euryarchaeota phylum accounted for 0–18% of the sequences in all samples, while the identified phylum Thermplasmatota exhibited 0 to 22% abundance (Fig. 6).

Bubble chart representing the percentage relative abundance of the 11 most abundant archaeal families. Each bubble represents % reads relative to all the major families in each sample. The first numbers (0, 16, and 49) represent the number of weeks, while the numbers after the hyphen (0.0, 0.5, 1.0, and 1.5) represent coagulant doses (0, 500, 1000, and 1500 ppm). R.A = Relative Abundance

Full size image

Sulfur cyclers

Most sequencing reads associated with sulfate- and sulfur-reducing microbes (comprising 0.3–24% of total sequence reads) were classified under the phylum Desulfobacterota (Fig. 7). Families linked to sulfate reduction, including Desulfosarcinaceae, Desulfobulbaceae, and Desulfatiglandaceae showed varied abundance across treatments. Sequences related to the Desulfosarcinaceae family comprised approximately 1% of reads in all alum treatments, except in the 500 ppm batch, where the abundance of reads increased from 7% (week 16) to 23% (week 49). In the ferric sulfate-treated samples, Desulfosarcinaceae sequences ranged from 0 to 4% in the controls but their initial abundance was higher in treated samples, followed by a gradual decline over time. In gypsum-treated batches, the proportion of Desulfosarcinaceae reads increased steadily, ranging from 2 to 15% throughout the experiment. Desulfobulbaceae and Desulfatiglandaceae families showed no consistent trends in any treatment and exhibited variable abundances between 1% and 4%. Sequences associated with microbes that conserve energy via S disproportionation [56, 57, 59] were identified and classified to the families Desulfocapsaceae and Desulurivibrionaceae. The relative abundance of Desulfocapsaceae family accounted for 1–20% across all samples with no particular patterns. In contrast, Desulfurivibrionaceae sequences remained consistent in gypsum-treated batches, ranging from 2 to 6% in all treatments. Additionally, sequences associated with sulfur reduction were classified under the SCADC1-2-3 (short-chain alkane–degrading culture) family of the Desufotomaculales order. The abundance of this family varied between 2% and 13% in all coagulant treatments.

Stacked bar plot showing % relative abundance of SO₄²⁻ and S reducers as a portion of the total reads in each sample. The first numbers (0, 16, and 49) represent the number of weeks, while the numbers after the hyphen (0.0, 0.5, 1.0, and 1.5) represent coagulant doses (0, 500, 1000, and 1500 ppm)

Full size image

Methanogens

The archaeal community was predominantly composed of well-known methanogens, particularly the genera Methanoregula, Methanolinea, and Methanosaeta (Fig. 8). These genera are involved in both hydrogenotrophic (Methanoregula and Methanolinea) and acetoclastic (Methanosaeta) methanogenesis pathways (Mohamad [52, 56, 78, 79].

Stacked bar plot representing % relative abundance of methanogens as a portion of the total reads in each sample. The first numbers (0, 16, and 49) represent the number of weeks, while the numbers after the hyphen (0.0, 0.5, 1.0, and 1.5) represent coagulant doses (0, 500, 1000, and 1500 ppm)

Full size image

The Methanoregula genus accounted for 20–46% of the total reads, with a maximum abundance observed in the 1500 ppm (week 49) ferric sulfate batch. The relative abundance of Methanolinea ranged from 3 to 27%, with the lowest abundance observed in the 1500 ppm (week 49) alum batch (Fig. 8). Reads assigned to genus Methanosaeta varied between 22% and 41%, with no general trends in the alum and ferric sulfate treatments; however, the percentage of reads decreased with time in all of the gypsum-treated batches (Fig. 8). Low abundances (3–18%) of reads associated with Candidatus Methanofastidiosum were also detected in all the samples (Fig. 8). Sequences corresponding to aerobic methanotrophs (genera Methylobacter and Methylocaldum) and anaerobic methane oxidizers (phyla NC10 and ANME-1 and 2) were not detected in any of the treatment batches.

Unrecovered light hydrocarbons added during bitumen froth treatment are the primary source of organic carbon for methanogenesis and other anaerobic respiration processes in tailings ponds [20, 70, 75, 79, 80]. The CH₄ production observed in the controls (no coagulant amendments) and the decreased SO₄²⁻ concentrations coupled with ΣH₂S production in the low coagulant dose amendments over time were indications of complex biogeochemical processes responsible for Fe, S, and C cycling in FFT deposits [80]. These processes are also important for greenhouse gas emissions, H₂S accumulation, and oxygen consumption in the FFT water cap above the tailings [18, 28, 59, 63, 79].

Sulfate-based coagulants stimulate microbial activities

The addition of sulfate-based coagulants supported decreased pH in all batches (Fig. 1). Larger pH decreases were associated with the addition of acidic alum and ferric sulfate solutions, while Al³⁺ or Fe³⁺ hydrolysis may also contribute to these observations. Changes in dissolved CO₂ concentrations may also influence pH in these batches. A slight pH decrease in the gypsum batches was likely driven by CO₂ production from anaerobic degradation and oxidation of organic carbon via microbial respiration [24, 32, 43, 49, 94]. The occurrence of carbon degradation was supported by the presence of known hydrocarbon degraders, SRB, and methanogens such as Anaerolineaceae, Desulfosarcinaceae, and Methanosaetaceae, which can produce CO₂ in anaerobic conditions [71]. The decreased pH also consumed alkalinity in all coagulant batches (Fig. 1), and low alkalinity can limit sulfate reduction by inhibiting SRB activity [13, 34]. This was reflected in the diversity and abundance indices, especially in alum and ferric sulfate treatments (Fig. 4).

Although SO₄²⁻ reduction has been reported in acidic environments, Fe³⁺ reduction tends to be thermodynamically favored at lower pH [12, 29, 41]. In the alum and ferric sulfate treatments, SO₄²⁻ addition corresponded to an initial increase in microbial diversity and abundance followed by a decrease over time as SO₄²⁻ concentrations decreased. The declining availability of labile organic carbon and electron acceptors (i.e., Fe³⁺ and SO₄²⁻) are among the factors that affect species dominance in FFT deposits [56]. The pore-water pH in gypsum treatments was circumneutral and the microbial diversity and abundance remained relatively constant (Fig. 4), which indicates that the energy available for microbial metabolism (usable energy) does not vary with pH [13]. The differences in diversity and abundance observed in the control batches of each coagulant could be a result of the heterogeneity of tailings ponds [56], which was also reflected in differences in the chemical composition of the initial samples (Figs. 1 and 2). These differences in initial chemical composition could influence the distribution of available growth substrates and microbial diversity in each sample.

Dissolved SO₄²⁻ concentrations decreased over time in all treatments except the 1000 and 1500 ppm alum batches (Fig. 2). Decreased SO₄²⁻ concentrations were also observed in laboratory experiments and zones below the TWI in FFT deposits and were attributed to microbial SO₄²⁻ reduction [18, 24, 71, 80, 93]. Sulfate supplied through the addition of SO₄²⁻-based coagulant served as the main electron acceptor for SRB during respiration [80]. The presence of known SRB such as Desulfosarcinaceae, Desulfobulbaceae, and Desulfatiglandaceae in the samples (Fig. 5.) supported SO₄²⁻ reduction and can also enrich FFT with elemental S [80, 93]. The initial increases observed in SO₄²⁻ concentrations in the alum and ferric sulfate treatments (Fig. 2) could be a result of the activities of Desulfocapsaceae and Desulfurivibrionaceae in the samples, which disproportionate S to produce SO₄²⁻ and H₂S [27, 56, 57, 59, 93]. These SRB use short-chain organic compounds or H₂ and depend on consortia of syntrophic microbes to biodegrade longer chain hydrocarbons to smaller molecules that are available for consumption [9, 14, 17, 28].

Identification and abundance of known hydrocarbon degraders such as Anaerolineaceae, Smithellaceae, Comamonadaceae, Rhodocyclaceae, Hydrogenophilaceae, Porticoccaceae, and Immundisolibacteraceae in all batches are indicative of microbial degradation of unrecovered hydrocarbon diluents [59, 65, 75, 83, 92]. The high SO₄²⁻ concentrations in all treated batches determined the dominant syntrophic processes in these experiments [59]. The SCADC1-2-3 family identified in the samples mainly has methanogenic or fermentative metabolism and can also incompletely oxidize methanogenic substrates [82, 83].

The ΣH₂S concentrations in the alum and ferric sulfate treatments were relatively low compared with the controls (Fig. 2), likely due to Fe²⁺ precipitation, which can react with H₂S and produce Fe(II) sulfide minerals [19, 24, 63, 73, 80]. Precipitated metal sulfides are a sink for the H₂S produced in the system, and this was supported by the decreased Fe concentration over time (Fig. 2). Sulfate concentrations in the 500 ppm ferric sulfate batch decreased sharply from week 1, with a corresponding decrease in ΣH₂S production. This can be attributed to the low microbial diversity and abundance over time (Fig. 4). The low abundance of SRB sequences, such as Desulfosarcinaceae, observed in this batch compared with other ferric sulfate treatments (Fig. 5) could indicate SO₄²⁻ consumption over time, which could restrict the activity of this family. This restriction could also result in less H₂S produced by SRB over time, which is also removed by secondary Fe(II) sulfide mineral precipitation. Gypsum-amended samples produced high ΣH₂S concentrations compared with alum and ferric sulfate treatments (Fig. 2), which could be related to low dissolved Fe in the samples (Fig. 2). This lack of dissolved Fe in the system could prevent Fe(II) sulfide mineral precipitation, resulting in eventual accumulation of H₂S (Fig. 9). Low Fe concentrations in the gypsum treatments could indicate that siderite, the main source of Fe in the system, did not dissolve at the high pH of these batches (Fig. 9). The microbial CO₂ production that drives down pH in gypsum-amended samples was not sufficient to result in a build-up of Fe in the system. The high ΣH₂S concentrations observed in the 1500 ppm gypsum batch could indicate total consumption of the limited Fe²⁺ in the system, which led to an uncontrolled accumulation of dissolved H₂S.

Plots of relationships between dissolved ΣH₂S and Fe in the alum, ferric, and gypsum batches

Full size image

The insoluble Fe(III) hydroxide formed during Fe³⁺ hydrolysis (Eq. 2) could serve as an electron acceptor for FeRB during respiration and produce Fe²⁺ [2, 18, 69, 70, 80]. The presence of known FeRB in the samples, which included the Comamonadaceae family [73, 91], also indicates Fe³⁺ reduction. The decrease in abundance of sequences related to the Comamonadaceae in the controls over time (Fig. 5) could indicate the consumption of Fe(III) sources by these microbes. This decrease could also be a result of competition with other hydrocarbon degraders such as Anaerolineaceae, which increased in abundance throughout the experiments. The abundance of reads related to the Comamonadaceae family was lower in the gypsum-treated batches than in the alum and ferric sulfate treatments (Fig. 5), which could indicate a limited source of Fe³⁺ as an electron acceptor during respiration. In the alum and gypsum treatments, where Fe³⁺ was not added, structural Fe(III) in FFT clay minerals or insoluble secondary goethite [α-FeO(OH)] and ferrihydrite [Fe(OH)₃] can also be used by FeRB to produce Fe²⁺ [31, 40, 73, 94]. The Geothermobacteraceae family, which was implicated in Fe³⁺ reduction [21, 48, 85], was present in low abundance in alum and ferric sulfate batches (≤ 2%), and was mostly not present in gypsum-treated samples. This low FeRB abundance could indicate restricted Fe³⁺ bioavailability. Iron(III) can also serve as an electron acceptor during microbial oxidation of NH₄⁺ to N₂ in anaerobic environments, a process known as anaerobic ammonium oxidation (anammox) [11].

Ammonium, which is the dominant N(-III) species at pH < 9.3, is important for maintaining water quality in FFT because NH₄⁺ oxidation can consume O₂ in the water cap [24, 36, 78]. The NH₄⁺ concentrations were low in both the controls and treated samples and the decrease in concentrations observed in the 1500 ppm alum batch could be dominated by ion exchange on the clay surfaces [51] rather than by anammox. This is supported by the absence of known anammox bacteria in the samples [89]; in addition, NO₃⁻ and NO₂⁻, which also serve as electron acceptors [54], were below detection in all batches. Moreover, anammox has not been widely reported in FFT but is well documented in freshwater sediments and organic matter–limited environments [22, 54, 64, 77].

Methanogenesis and organic carbon degradation

Methanogenic Archaea have been identified as the major CH₄ producers in anaerobic environments [36, 37, 56, 59, 79]. The increasing concentrations of dissolved CH₄ observed in the control experiments indicate the potential for methanogenesis (Fig. 3.). Despite the presence of methanogens, CH₄ concentrations were below detection at week 0, which could represent the lag time for methanogens to become active after exposure to O₂ during the experimental setup under ambient laboratory conditions. The dominant archaeal taxa included acetoclastic Methanosaeta and two hydrogenotrophic methanogens from Methanoregula and Methanolinea. These taxa are consistent with the dominant methanogens previously observed within 6 to 30 m below surface in MLSB by [56]. These authors identified Methanosaeta, which uses acetate, as the dominant methanogenic archaeal taxa in MLSB. However, [78] and 79 observed hydrogenotrophic methanogens, which use CO₂/H₂, as the major drivers of methanogenesis, in their laboratory experiments with samples collected from other active tailings ponds. This dominance of hydrogenotrophic methanogenesis could also be aided by H₂ present in the anaerobic chamber, which can serve as the electron donor for hydrogenotrophic methanogens.

In the controls, there was an increase in Methanosaeta and decrease in Methanoregula abundance (Fig. 6), which corresponded with an increase in abundance of Anaerolineaceae, decrease in Comamonadaceae, and low abundance of other known hydrocarbon degraders such as Smithellaceae, Rhodocyclaceae, Hydrogenophilaceae, and Porticoccaceae (Fig. 5, [28, 59, 66, 83]. This could indicate that the activities of acetoclastic methanogens increased as these syntrophic microbes anaerobically degraded long-chain hydrocarbons to provide additional acetate (CH₃COO⁻) used by acetoclastic methanogens. However, hydrogenotrophic methanogens can also consume the H₂ and CO₂ produced in the fermentation process [71, 78, 79, 84]. Despite the decreased abundance of sequences related to hydrogenotrophic methanogens in the controls over time, they still represented 46% of the total reads, which could indicate that abundant CO₂ was produced by syntrophic microbes and oxidation of acetate and was available to H₂-consuming methanogens [79].

The abundance of these methanogenic Archaea in the samples increased CH₄ concentrations to > 1.00 mg L⁻¹ in all control batches at week 8 (Figs. 3 and 6); the differences in concentrations observed in these controls could indicate geochemical heterogeneity within tailings ponds, which can affect microbial activity [28]. Methane concentrations were relatively low with coagulant addition in all treatments, except in the 500 ppm alum batch, where CH₄ concentrations increased from week 16 to the end of the experiment (Fig. 3). This increase correlated with decrease in SO₄²⁻ concentrations in the sample (Fig. 2), which supported the inhibition of methanogenesis by SO₄²⁻ reduction [26, 36, 59, 63].

Interactions among biogeochemical redox processes

The presence of FeRB, SRB, and methanogens in our experiments is indicative of complex biogeochemical redox cycling in treated FFT. [28] suggested that co-occurrence of these microbial groups reflects abundant substrate production via syngenetic anaerobic hydrocarbon degradation. The SO₄²⁻ and Fe³⁺ contribution from coagulants, coupled with growth substrates and electron donors from naphtha diluents and fermentation products of syntrophic degraders, imply that electron donors and acceptors were not limiting. Consequently, these major biogeochemical processes could take place simultaneously (Lovely and Philip 1987; [80]. For example, the abundant hydrocarbon degraders observed in these samples (Fig. 5) will produce labile organic carbon for other microbes [59]. Sulfate reducers can use this labile organic carbon and other short-chain compounds present in hydrocarbon diluents, or H₂, during respiration to produce H₂S and eventually CO₂ [78]. This syntrophic breakdown of organic carbon to CO₂ can stimulate hydrogenotrophic methanogenesis [79]. Acetoclastic methanogens can also use acetate to produce CH₄ and CO₂, which can be used by hydrogenotrophic methanogens [56]. Iron reducers can use CH₄ as a growth substrate and an electron donor during Fe³⁺ reduction [6, 45]. Iron(II) sulfide or FeS₂ precipitation by a combined effort of FeRB and SRB will enhance the mutual relationship between these microbes [13, 41, 45, 53, 75].

The production of Fe²⁺, H₂S, and CH₄ in these batch experiments indicates the co-existence of these microbes in FFT, but the activity of one microbe might impact the abundance of others. We observed that the addition of sulfate-based coagulants suppressed methanogenesis in most batches (Fig. 3). The inhibitory relationships between SRB and methanogens have previously been observed below the TWI and in the laboratory [26, 36, 59, 63]. The presence of SO₄²⁻ in FFT pore water can give SRB a thermodynamic and kinetic advantage over methanogens when competing for the same fermentation products (acetate and H₂) [35, 36, 63], which is supported by the observed increase in CH₄ concentrations following SO₄²⁻ depletion in the 500 ppm ferric sulfate batch (Fig. 3) and negative correlations between SO₄²⁻ and CH₄ concentrations in all 500 ppm treatments (Supplementary Material, Table A5).

Implications for FFT biogeochemistry and management

Methanogenesis has been linked to accelerated FFT dewatering, yet biogenic CH₄ contributes significantly to greenhouse gas emissions from oil sands tailings ponds [72, 79] He at al., 2024). Interactions among microbial processes influence both water quality and gas emissions, which have important implications for management and reclamation of oil sands tailings [60, 67, 79, 93]. Current results are consistent with past studies reporting similar trends in CH₄ production in the presence of sulfate. [59] observed a ~ 50% decrease in CH₄ production in FFT containing 2 mM SO₄²⁻, while [79] predicted that more than 5 × 10⁶ L day⁻¹ of CH₄ production could be prevented by stimulating SRB activity.

Methane exsolution and ebullition is a key driver of CH₄ emissions from FTT deposits [30, 43]. Unlike dissolved fluxes, ebullition facilitates rapid upward transport of CH₄ gas from FFT and limits potential for methanotrophs to oxidize CH₄ to CO₂ in overlying water [3, 30]. Although this process can limit dissolved CH₄ fluxes, corresponding O₂ consumption can contribute to anoxia in tailings ponds or pit lakes [8, 62]. Excess H₂S production in treated FFT also presents challenges for water quality and gas emissions from FFT deposits. Like CH₄, oxidation of dissolved H₂S consumes O₂ and can contribute to anoxia [8, 62], while ebullition also contribute H₂S emissions [74]. Elevated H₂S concentrations can directly impact water quality and further complicate aquatic reclamation [60, 61, 78]. Our results are consistent with past studies reporting that Fe(II)-sulfide precipitation can limit H₂S accumulation in FFT pore water when Fe(II) availability exceeds the H₂S production capacity [18, 63].

Complex interactions among hydrocarbon degraders, sulfate-reducing bacteria, iron reducing bacteria, and methanogens mediate biogenic gas production and pore-water chemistry in treated FFT. Our results show that sulfate-based coagulants stimulate microbial sulfate reduction and suppress methanogenesis in treated oil sands FFT compared to controls. Methane production resumed following sulfate depletion in the gypsum treatment with the lowest dose (i.e., 500 ppm), whereas sulfate depletion was not observed in other batches. The addition of alum and ferric coagulations produced notable pH decreases attributed to the acidic solutions, Al³⁺ and Fe³⁺ hydrolysis, and associated CO₂ production. These experiments also exhibited distinct changes in microbial community structure, which are attributed to the observed pH changes. Accumulation of dissolved H₂S generated during sulfate reduction was limited by Fe(II) sulfide precipitation, except at the highest gypsum dose (i.e., 1500 ppm) where H₂S production exceeded Fe(II) availability.

This study shows that the addition of sulfate-based coagulants has substantial implications for biogeochemical cycling of carbon, sulfur, and iron in treated FFT deposits. Moreover, our results reveal the intricate balance between these processes and their implications for biogenic gas production and pore-water chemistry. Specifically, suppression of methane production in treated FFT can be temporary, with sulfate depletion preceding methane production if labile organic carbon remains available. Additionally, depletion of available Fe(II) via sulfide-mineral precipitation may be followed by increased H2S concentrations under sustained sulfate-reducing conditions. These observations suggest that methane emissions can be suppressed or, potentially, mitigated in treated FFT where (i) sulfate addition stimulates robust sulfate reduction, (ii) iron availability exceeds H₂S production capacity, and (iii) labile organic carbon is depleted via sulfate and iron reduction. Achieving this balance among biogeochemical redox processes could enhance environmental outcomes of both aquatic and terrestrial FFT reclamation.

Sequence data that support the findings of this study have been deposited European Molecular Biology Laboratory-European Bioinformatics Institute (EMBL-EBI) BioStudies Database with the primary accession code S‑BSST1350. All additional data included in this study are presented.European Nucleotide Archive with the primary accession code PRJWB13140.

This research was supported by the Towards Environmentally Responsible Resource Extraction Network (TERRE-NET), which was established through the Natural Sciences and Engineering Research Council of Canada (NSERC) Strategic Partnership Grant for Networks program (Grant No. NETGP-479708-2015). We thank Syncrude Canada Ltd. for providing the fluid fine tailings. We also thank Dr. Jing Chen, Mattea Cowell, Noel Galuschik, and Dr. Joyce McBeth for their assistance.

Authors and Affiliations

1. Department of Geological Sciences, University of Saskatchewan, 114 Science Place, Saskatoon, SK, S7N 5E2, Canada

   Philip A. Adene, Mojtaba Abdolahnezhad & Matthew B. J. Lindsay

2. Department of Civil & Environmental Engineering, University of Alberta, Alberta, Edmonton, AB, T6G 1H9, Canada

   Mian N. Anwar & Ania C. Ulrich

Contributions

P.A.A. performed the experiments; P.A.A. and M.A. analyzed the samples; P.A.A. and M.N.A. analyzed the data; P.A., M.N.A., and M.B.J.L. interpreted the data; P.A.A. prepared the tables and figures, P.A.A. drafted the original manuscript; P.A.A., M.N.A., A.C.U., and M.B.J.L. reviewed and edited the manuscript.

Corresponding author

Correspondence to Matthew B. J. Lindsay.

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions